first dataset version

data/emodb.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e10f3ddf042f4ab930073b2dcf55c77f04bd7b4b57cb4f990d815a35291be04b
+size 7463

data/wav.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7e28cbd561c22303b55bac2e1190199afc7aa3773498f94154466877b06033e0
+size 39054275

emodb_enriched.py

@@ -0,0 +1,93 @@
+
+
+import datasets
+
+import pandas as pd
+
+from pathlib import Path
+from datasets import load_dataset
+
+
+_DATA_URLS = {
+'raw': "data/wav.zip",
+'metadata': 'data/emodb.parquet'
+}
+
+_GENDER = ['male', 'female']
+_EMOTION = ['happiness', 'neutral', 'anger', 'boredom', 'disgust', 'fear', 'sadness']
+
+
+class emoDBEnriched(datasets.GeneratorBasedBuilder):
+    """EmoDBEnriched Data Set"""
+
+    BUILDER_CONFIGS = [
+        datasets.BuilderConfig(
+            name="enriched",
+            version=datasets.Version("1.0.0", ""),
+            description="Import of enriched EmoDB Data Set",
+        )
+    ]
+
+    def _info(self):
+        return datasets.DatasetInfo(           
+            features=datasets.Features(
+                {
+                    "audio": datasets.Value("string"),                   
+                    "emotion": datasets.features.ClassLabel(names=_EMOTION),
+                    "gender": datasets.features.ClassLabel(names=_GENDER),
+                    "emotion_str": datasets.Value("string"),                   
+                    "gender_str": datasets.Value("string"), 
+                    "age": datasets.Value("int32")
+                }
+            ),           
+        )
+
+    def _split_generators(self, dl_manager):
+        
+        archive_path = dl_manager.download_and_extract(_DATA_URLS['raw'])
+        metadata = pd.read_parquet(dl_manager.download(_DATA_URLS['metadata']))        
+    
+
+        return [
+            datasets.SplitGenerator(
+                name=datasets.Split.TRAIN, gen_kwargs={"archive_path": archive_path, 'metadata': metadata, "split": "train"}
+            ),
+            
+        ]
+
+    def _generate_examples(self, archive_path, metadata, split):
+        """This function returns the examples in the raw (text) form."""
+
+        if split == "train":
+            df = metadata
+
+
+        if split == "all":
+            df = metadata      
+
+        
+
+       
+        for index, row in df.iterrows():
+            audio_path = archive_path + "/" + row['file']
+            #img = {"path": img_path, "bytes": None}
+            #print(str(len(row['probabilities'])))
+            #print(str(index))
+            result = {
+                "audio": audio_path,                   
+                "emotion": row['emotion'],
+                "gender": row['gender'],
+                "emotion_str": row['emotion'],                   
+                "gender_str": row['gender'], 
+                "age": row['age'],             
+            }
+            yield index, result
+
+       
+
+
+
+
+
+#if __name__ == "__main__":
+#    ds = load_dataset("emodb_enriched.py", split="all")